One acknowledgement, two possible FR5 attempts

Only the label. Local counter resets too.

Then show both possible matches rather than picking the nearer timestamp. What order can you establish within each individual source?

No offsets measured then. We only have the printed times.

Python shows timeout, restart, then receipt. Can't tell whether the response belongs to the earlier send.

Mark that receipt as unmatched in the incident account. Don't let the report quietly award it to the retry because it appears after the restart.

I've seen 'retry passed' get written on a shift board while the engineer was still deciding which result was which. Kai, what have the operators been told about this bracket?

That needs the quality owner's decision while the attempt remains uncertain. The software history should not silently decide the bracket's status from this unmatched response.

Quality has it on hold. Shift note now says uncertain result, not retry passed.

For future attempts, ask the interface owner about identities that survive restarts and are echoed in the response. The local counter you have cannot make that link reliably.

And agree the clock recording scheme for future diagnostics. Better timestamps would help the timeline, but you will still need that request-response identity to distinguish delayed replies.

Both changes proposed. Old incident stays unmatched; no attempt magically completed by sorting the spreadsheet.

Thanks for asking about the shift note, Leah. That was getting ahead of the evidence.

Try the revised wording on someone who wasn't in this discussion. Ask what they'd do with the bracket. You'll find out whether 'uncertain' actually leads them to the right person.